PeopleofSuperintelligence

Researcher

Georg Ostrovski

Georg Ostrovski is an AI researcher whose work spans game-theoretic dynamics and deep reinforcement learning. He published a study of maps inspired by game theory in 2013. His research includes PixelCNN exploration for Atari, quantile methods for reinforcement learning and image generation, and research on learning from observational data. In August 2025, he contributed insights and support to Genie 3; as of October 2026, he leads DeepMind’s Reinforcement Learning Engineering team, serves on Amfileon’s Scientific Board and holds a mathematics PhD from the University of Warwick.

Now at
Google DeepMind and Amfileon
Works
33
Citations
36,064
h-index
14
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2013

    Maps inspired by game theory

    In 2013, he presents a one-parameter family of continuous, piecewise-affine, area-preserving maps of the square inspired by a dynamical system in game theory.

  2. 2017

    PixelCNN for Atari exploration

    In 2017, he coauthors Count-Based Exploration with Neural Density Models, which uses PixelCNN pseudo-counts with reinforcement-learning agents and improves results on several hard Atari games.

  3. 2018

    Quantile networks for games and images

    In 2018, he coauthors Implicit Quantile Networks for Distributional Reinforcement Learning, which uses quantile regression to approximate return distributions and improves performance on the 57 Atari games. He also coauthors Autoregressive Quantile Networks for Generative Modeling, which extends PixelCNN with quantile regression and reports image-modeling results on CIFAR-10 and ImageNet.

  4. 2021

    Passive learning in deep RL

    At NeurIPS in 2021, he coauthors The Difficulty of Passive Learning in Deep Reinforcement Learning with Pablo Samuel Castro and Will Dabney. The paper identifies function approximation and fixed data distributions as key factors in learning from observational data.

  5. 2023

    Quantile credit assignment

    In 2023, he coauthors Quantile Credit Assignment, which uses return quantiles to separate environmental randomness from the controllable effects of an agent’s actions.

  6. 2024

    Quantile TD convergence

    In 2024, he coauthors An Analysis of Quantile Temporal-Difference Learning, which proves convergence to the fixed points of a related family of dynamic programming procedures.

  7. August 2025

    Insights for Genie 3

    In August 2025, he contributes insights and support to Genie 3, Google DeepMind’s general-purpose world model for interactive environments.

  8. As of October 2026

    Reinforcement learning engineering lead

    As of October 2026, he leads DeepMind’s Reinforcement Learning Engineering team and serves on Amfileon’s Scientific Board. He holds a mathematics PhD from the University of Warwick.

Career

  1. Since 2015Google DeepMindTeam Lead
  2. NowAmfileonScientific Board member
  3. 2016–2023Google

Education

  1. University of WarwickPhD in Mathematics

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Georg Ostrovski. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Georg Ostrovski yet. Send us a free-licensed photo.

  1. Dynamics of a Continuous Piecewise Affine Map of the Square arxiv.org. Checked 22 October 2026.
  2. DeepMind papers at ICML 2017 (part three) — Google DeepMind deepmind.google. Checked 22 October 2026.
  3. Count-Based Exploration with Neural Density Models proceedings.mlr.press. Checked 22 October 2026.
  4. DeepMind papers at ICML 2018 — Google DeepMind deepmind.google. Checked 22 October 2026.
  5. Implicit Quantile Networks for Distributional Reinforcement Learning proceedings.mlr.press. Checked 22 October 2026.
  6. Autoregressive Quantile Networks for Generative Modeling proceedings.mlr.press. Checked 22 October 2026.
  7. The Difficulty of Passive Learning in Deep Reinforcement Learning proceedings.neurips.cc. Checked 22 October 2026.
  8. Quantile Credit Assignment proceedings.mlr.press. Checked 22 October 2026.
  9. An Analysis of Quantile Temporal-Difference Learning jmlr.org. Checked 22 October 2026.
  10. Genie 3: A new frontier for world models — Google DeepMind deepmind.google. Checked 22 October 2026.
  11. When markets break, this strategy accelerates - inside Amfileon's Crisis-Alpha machine — Opalesque opalesque.com. Checked 22 October 2026.
  12. Home | Amfileon amfileon.com. Checked 22 October 2026.